JOURNAL OF GEOPHYSICAL RESEARCH, YOLo 96, NO. C7, PAGES 12,667-12,683, JULY IS, 1991 The SeasonalCirculation of the Upper Oceanin the Bay of Bengal JAMEST. POTEMRA,IMARK E. LUTHER,2AND JAMESJ. O'BRIEN MesoscaleAir-Sea Interaction Group,Florida State University, Tallahassee Analysis of the results of a multilayer, adiabatic, numerical model of the upper Indian Ocean, driven by climatological monthly mean winds, shows that the simulated currents in the northeastern Indian Ocean are in general agreement with available observations and interpretations. The main features of the ocean currents include large anticyclonic flow in the Bay of Bengal surface waters during the northern hemisphere winter. This gyre decays into eddies in spring and then transitions into a weaker, cyclonic gyre by late summer. The western recirculation region of this flow is an intensified western boundary current which changesdirection twice during the year. In the Andaman Sea, east of the Bay of Bengal, the oceanic flow changesdirection twice during the year; it is cyclonic during the spring and early summer and anticyclonic the rest of the year. Flow in the equatorial region shows the North Equatorial Current (NBC) flowing west during winter. Further south is the eastward flowing Equatorial Counter Current (ECC) and the westward flowing South Equatorial Current. In summer, the NEC switches direction, joins the ECC, and forms the Indian Monsoon Current. Investigation of the second layer of the model (the upper 450 m of the ocean) shows that flow during much of the year is baroclinic (strong vertical shear). Model layer thickness reveals coastal Kelvin waves propa!~ting along the coast, traveling the entire perimeter of the Andaman Sea and the Bay of Bengal. This wave excites westward propagating Rossby waves into the interior of the bay. Time series analysis of transport calculations yield significant peaks in the 20- to 3o-day range and 50- to ro-day range which are not likely directly forced by the applied wind stress. INTRODUCTION few excellentatlases [see Duing, 1970;Wyrtki, 1971;Has- tenrath and Lamb, 1979;Cutler and Swallow, 1984].Al- The Indian Ocean is a particularly interesting, dynamically though these sourcescontribute significantlyto the under- rich area because of the changing wind patterns associated standingof dynamicsin the tropical Indian Ocean,data on with the Indian Monsoon. In spite of this, it is a relatively smallerscales, particularly in the Bay of Bengal,are some- poorly studied area. In addition, data coverage for the area is what limited. sparse. While some data sets do exist, they are predomi- The focus of this study therefore is to use a realistic, nantly on large space and time scales. Recent efforts, such as wind-drivenmodel to simulateocean currents in the region the monsoon experiment (MONEX) of the Global Atmo- of the Bay of Bengal,from 75°E to 111°Eand from 70Sto spheric Research Program (GARP), the First GARP Global 23°N. Similar studieshave beendone for other areasof the Experiment (FGGE), and the Indian Ocean Experiment Indian Ocean [Jensen, 1990] and using different models (INDEX), have substantially improved the availability of [Woodberryet al., 1989].A review of modelingefforts in the measured data in this region. Indian Oceanis given by Luther [1987].The results of the Individual studies have also contributed in recent years to model are comparedwith measureddata in an attempt to the understanding of the dynamics in the Indian Ocean. An understandthe dynamicsof the area.The modelused for this early effort by Wyrtki [1961] synthesized the available data study, developedat the Florida State University [Jensen, sets (at the time) of the properties of the Southeast Asian 1990],is a reducedgravity, three and one-halflayer model. waters. More recently, Legeckis [1987] demonstrated the Sincethe model has multiple layers, the subsurfacecur- appearance of a western boundary current in the Bay of rents can be studiedas well as the surface.The lowest layer Bengal using sea surface temperature data obtained from the is at rest, however, and only the results from the first advanced very high resolution radiometer (A VHRR) carried (surface)and second(subsurface) layers will be examined on board the NOAA 9 satellite. Data coverage from this here. satellite encompassesthe entire region of the Bay of Bengal Figure 1 showsthe typical large-scalecirculation pattern [McClain et al., 1985]. Rao et al. [1989] have generated associatedwith the winter and summermonsoons as com- mean monthly mixed layer depth, sea surface temperature, piled in Monatskartenfiir den IndischenOzean (Deutsches and surface current climatologies for the tropical Indian HydrographischesInstitut [1960], as depicted by Diiing Ocean. Molinari et al. [1960] analyzed three different sets of [1970]). The general surface circulation of the region in satellite-tracked drifting buoys and compiled a monthly northernwinter includeslarge anticyclonic flow in the Bay of climatology of surface currents in the tropical Indian Ocean. Bengal, a westward flowing North Equatorial Current Finally, oceanic data for the Indian Ocean is compiled in a (NEC), and eastward flowing Equatorial Counter Current (ECC) and a westward flowing South Equatorial Current I Now at STX, NASA Goddard Space Flight Center, Greenbelt, (SEC). The northern summer pattern is characterizedby M~land. counterrotatingeddies in the Bay of Bengal[Duing, 1970], 2Now at Department of Marine Science, University of South an eastwardflowing Indian MonsoonCurrent (IMC) and a Florida, St. Petersburg,. westwardflowing SEC. Copyright 1991 by the American Geophysical Union. The model reproducesthese flows, in generalagreement with the atlasespreviously discussed.In this presentation Paper number 91JCOI045. 0148-0227/91/9IJC-O 1 045$05.00 the model is first described. Next, a description of the 12.'667 12,668 PoTEMRA ET AL UPPER OcEAN CIRCULAnON IN BAY OF BENGAL 10° .to' 60. 80. 100. 120' 140"£ a " 20°1 ~...~ - 1-'~, ~", ( 20. " .. " 4- ::' I~ .- ..,J Il'v .-~.- O' ... ... ..- ..- \.~"\ -- -..~, ""-~~-. -- ... -~---. - ..- . O' +- -.- '*"" 4- ~ J .-.;. ~.'~;;. ..A ~ ~ +-- 4-- ~\, ~~?J"'~.~~ 20° ~, +- ~-' 4-' ~.....- ,"", 120" ..,:-.1 '.."- .'~'" ( ~ ' -, A. ~ . -- -- ~ ...' ~ --. ' ~ ¥ ~ , c... -+ -.- --+ ~' I - C-+-" --. ..", ./-"' -+ / + '- '..- .. '-', 40. +~-+ + ,~ 40. -+ - ~-+-+ ~ ~ '" '~.::+~I " ,-" . ..-+ ~, ~ -:-.. 20. 40' 60. 80. 100. 120. 140.£ Fig. 1. Large-scale circulation as observed during (a) the winter (northeast) monsoon, and (b) the summer (southwest) monsoon. (Deutsche.r Hydrographisches Institut [1960], courtesy of W. Diling). climatology of the winds that force the model is given. work of Luther and O'Brien. This model, however, is Finally, the resultsof the modelare presented,first in terms composed of four isopycnal layers. The advantage of a of generalcirculation patternsand later in terms of oscilla- multilayered model is the ability of studying both surface and tions of certain features. subsurface currents. In addition, multiple layers provide greater accuracy in representing low vertical modes. 2. MODEL The layers of this model are assumed to have positive thickness everywhere at all times. In other words, the layers Isopycnic layered models have proven to be useful in the can not merge or surface, and the bottom topography is study of ocean dynamics. The work of Luther and O'Brien always in the lowest layer. The vertical modes for the Indian [1985] and Luther et al. [1985] demonstrates how a reduced Ocean, given by Gent et al. [1983], are used to select the gravity model can correctly simulate the currents of the initial layer thicknesses and densities. The four densities are western Indian Ocean. Further, Woodberry et al. [1989] chosen to be 1023.9, 1026.2, 1027.3,and 1027.9kg m -3. The show this model to be equally capable of reproducing the initial thicknesses of the layers are taken as 200 m for the top complex current system of the southern Indian Ocean. The layer, 250 m for the second layer, and 400 m for the third. success of these efforts gives rise to the numerical model For this model, horizontal eddy viscosity is based on the used for this investigation of currents in the Bay of Ben~. velocity field so that a thin layer with large velocities would The model employed for this study was developed at the experience more friction than a thick layer with equivalent Florida State University by Jensen [1990] and is based on the transport [Jensen, 1990]. The magnitude of the horizontal PoTEdA ET AL.: UPPEa OcEAN CIRCULADON IN BAY OF BENOAL 12,669 i-I w~=o Hl>Hmin ap.+ PJI--a" 9 ""~ (Pi - PI> a~ a. I-I For the bottom two layers this term is set to zero. The ~] effect of this entrainment on the upper layer momentum balance is neglected. Incorporating this entrainment velocity + (-"/,, - (1) only in the continuity equation represents turbulent entrain- Pi ment where entrained water into the upper layer has zero velocity and is warmed instantaneously to the upper layer density. For this particular study, however, this adjustment is rarely active. The value for H minis taken to be 60 m and the time constant 1"~ is 1.2 hours. j-1 The boundary conditions for this set of equations are given a." -g 2 (PJ- PI) ~ in two parts. First, for closed boundaries, i.e., land, the I-I as] no-slip condition is incorporated. At coastal points, there- fore, both components of the transport are set to zero. Next, + ( oj",'0' 0 " for open boundaries, a Sommerfield radiation condition is ~ i +F' (2) applied.